26 
Mr. Christie on the effects of temperature on 
quently by using these, the value of ^ — P foi* the de- 
termination of cos.* ((p + A (?>) would be true to the 6th 
figure, which would give cp + A (p to the tenth of a second. 
Such a degree of accuracy may appear quite uncalled for by 
the nature of the observations, but from the manner which 
I adopted for correcting them, it was necessary to guard 
against any accumulation of error. 
From what I have said, we have 
cos.“ ((p + A (p) = ^ I y 
but 
Y = p + Q cos.- (f>, 
and therefore, 
cos.^ ((p ^ tp) = { Q <P + I - (G) 
Having, as we have seen, determined by observation A F 
in terms of M, and y, being computed from the equation (C), 
for any angle cp, the value of A cp would be readily computed 
from this formula : that is, we could obtain from it the cor- 
rection to be made in any observed angle, for a change of i® 
in the temperature of the magnets, whether that tempera- 
ture were increasing or decreasing, only observing that A F 
is minus for an increase of temperature, and plus for a 
decrease. 
The method which I have adopted for reducing the ob- 
served values of (p to what they would have been, had the 
temperature of the magnets been constant, is this : the 
observed values of cp being comprised between 74° and 86®, 
I computed the values of A (p, both plus and minus, at inter- 
